add cpp11poppler to suggests

Author: pachadotdev

DESCRIPTION

@@ -39,7 +39,10 @@ Suggests:
     knitr,
     tibble,
     rmarkdown,
-    testthat (>= 3.0.0)
+    testthat (>= 3.0.0),
+    cpp11poppler
+Remotes: 
+    pachadotdev/cpp11poppler
 Encoding: UTF-8
 VignetteBuilder: knitr
 Language: en-US

R/cpp11.R

@@ -47,15 +47,3 @@ ocr_raw_data <- function(input, ptr) {
 ocr_file_data <- function(file, ptr) {
   .Call(`_cpp11tesseract_ocr_file_data`, file, ptr)
 }
-
-n_pages <- function(file_path, opw, upw) {
-  .Call(`_cpp11tesseract_n_pages`, file_path, opw, upw)
-}
-
-get_poppler_config <- function() {
-  .Call(`_cpp11tesseract_get_poppler_config`)
-}
-
-poppler_convert <- function(file_path, format, pages, names, dpi, opw, upw, antialiasing, text_antialiasing) {
-  .Call(`_cpp11tesseract_poppler_convert`, file_path, format, pages, names, dpi, opw, upw, antialiasing, text_antialiasing)
-}

R/images.R

@@ -1,46 +1,7 @@
-is.pdf <- function(x) {
-  grepl("\\.pdf$", tolower(x))
-}
-
 is.tiff <- function(x) {
   grepl("\\.tiff$", tolower(x))
 }
 
-pdf_convert <- function(pdf, format = "png", pages = NULL, dpi = 72,
-                        antialias = TRUE, opw = "", upw = "") {
-  config <- get_poppler_config()
-  
-  if (isFALSE(config$render) || isFALSE(length(config$format) > 0)) {
-    stop("You version of libppoppler does not support rendering")
-  }
-  
-  format <- match.arg(format, config$format)
-  
-  if (is.null(pages)) {
-    pages <- seq_len(n_pages(file, opw = opw, upw = upw))
-  }
-
-  if (isFALSE(is.numeric(pages)) || isFALSE(length(pages) > 0)) {
-    stop("Argument 'pages' must be a one-indexed vector of page numbers")
-  }
-  
-  antialiasing <- isTRUE(antialias) || isTRUE(antialias == "draw")
-
-  text_antialiasing <- isTRUE(antialias) || isTRUE(antialias == "text")
-
-  dout <- tempdir()
-  suppressWarnings(try(dir.create(dout)))
-
-  filenames <- file.path(dout, sprintf("%s-%03d.%s",
-    # basename(pdf),
-    # remove the file extension
-    tools::file_path_sans_ext(basename(pdf)),
-    pages, format)
-  )
-
-  poppler_convert(pdf, format, pages, filenames, dpi, opw, upw, antialiasing, text_antialiasing)
-}
-
 tiff_convert <- function(tiff, format = "png", dpi = 72) {
   stopifnot(requireNamespace("magick", quietly = TRUE))
   magick::image_convert(magick::image_read(tiff), format = format)

R/ocr.R

@@ -41,21 +41,14 @@ ocr <- function(file, engine = tesseract("eng"), HOCR = FALSE, opw = "", upw = "
       magick::image_write(x, tmp, format = "PNG", density = "300x300")
       ocr(tmp, engine = engine, HOCR = HOCR)
     }, character(1))
-  } else if (isTRUE(is.character(file)) && isFALSE(is.pdf(file))) {
+  } else if (isTRUE(is.character(file))) {
     if (isFALSE(is.tiff(file))) {
       vapply(file, ocr_file, character(1), ptr = engine, HOCR = HOCR, USE.NAMES = FALSE)
     } else {
       ocr(tiff_convert(file), engine, HOCR = HOCR)
     }
   } else if (isTRUE(is.raw(file))) {
     ocr_raw(file, engine, HOCR = HOCR)
-  } else if (isTRUE(is.pdf(file))) {
-    n <- n_pages(file, opw = opw, upw = upw)
-    fout <- pdf_convert(file, format = "png", pages = 1:n, opw = opw, upw = upw)
-    out <- vapply(fout, function(x) ocr(x, engine = engine, HOCR = HOCR), character(1))
-    unlink(fout)
-    names(out) <- NULL
-    out
   } else {
     stop("Argument 'file' must be file-path, url or raw vector")
   }

src/cpp11.cpp

@@ -89,39 +89,15 @@ extern "C" SEXP _cpp11tesseract_ocr_file_data(SEXP file, SEXP ptr) {
     return cpp11::as_sexp(ocr_file_data(cpp11::as_cpp<cpp11::decay_t<const std::string &>>(file), cpp11::as_cpp<cpp11::decay_t<TessPtr>>(ptr)));
   END_CPP11
 }
-// cpp11tesseract.cpp
-int n_pages(const std::string & file_path, const std::string & opw, const std::string & upw);
-extern "C" SEXP _cpp11tesseract_n_pages(SEXP file_path, SEXP opw, SEXP upw) {
-  BEGIN_CPP11
-    return cpp11::as_sexp(n_pages(cpp11::as_cpp<cpp11::decay_t<const std::string &>>(file_path), cpp11::as_cpp<cpp11::decay_t<const std::string &>>(opw), cpp11::as_cpp<cpp11::decay_t<const std::string &>>(upw)));
-  END_CPP11
-}
-// cpp11tesseract.cpp
-list get_poppler_config();
-extern "C" SEXP _cpp11tesseract_get_poppler_config() {
-  BEGIN_CPP11
-    return cpp11::as_sexp(get_poppler_config());
-  END_CPP11
-}
-// cpp11tesseract.cpp
-std::vector<std::string> poppler_convert(const std::string & file_path, const std::string & format, const std::vector<int> & pages, const std::vector<std::string> & names, const double & dpi, const std::string & opw, const std::string & upw, const bool & antialiasing, const bool & text_antialiasing);
-extern "C" SEXP _cpp11tesseract_poppler_convert(SEXP file_path, SEXP format, SEXP pages, SEXP names, SEXP dpi, SEXP opw, SEXP upw, SEXP antialiasing, SEXP text_antialiasing) {
-  BEGIN_CPP11
-    return cpp11::as_sexp(poppler_convert(cpp11::as_cpp<cpp11::decay_t<const std::string &>>(file_path), cpp11::as_cpp<cpp11::decay_t<const std::string &>>(format), cpp11::as_cpp<cpp11::decay_t<const std::vector<int> &>>(pages), cpp11::as_cpp<cpp11::decay_t<const std::vector<std::string> &>>(names), cpp11::as_cpp<cpp11::decay_t<const double &>>(dpi), cpp11::as_cpp<cpp11::decay_t<const std::string &>>(opw), cpp11::as_cpp<cpp11::decay_t<const std::string &>>(upw), cpp11::as_cpp<cpp11::decay_t<const bool &>>(antialiasing), cpp11::as_cpp<cpp11::decay_t<const bool &>>(text_antialiasing)));
-  END_CPP11
-}
 
 extern "C" {
 static const R_CallMethodDef CallEntries[] = {
     {"_cpp11tesseract_engine_info_internal",          (DL_FUNC) &_cpp11tesseract_engine_info_internal,          1},
     {"_cpp11tesseract_get_param_values",              (DL_FUNC) &_cpp11tesseract_get_param_values,              2},
-    {"_cpp11tesseract_get_poppler_config",            (DL_FUNC) &_cpp11tesseract_get_poppler_config,            0},
-    {"_cpp11tesseract_n_pages",                       (DL_FUNC) &_cpp11tesseract_n_pages,                       3},
     {"_cpp11tesseract_ocr_file",                      (DL_FUNC) &_cpp11tesseract_ocr_file,                      3},
     {"_cpp11tesseract_ocr_file_data",                 (DL_FUNC) &_cpp11tesseract_ocr_file_data,                 2},
     {"_cpp11tesseract_ocr_raw",                       (DL_FUNC) &_cpp11tesseract_ocr_raw,                       3},
     {"_cpp11tesseract_ocr_raw_data",                  (DL_FUNC) &_cpp11tesseract_ocr_raw_data,                  2},
-    {"_cpp11tesseract_poppler_convert",               (DL_FUNC) &_cpp11tesseract_poppler_convert,               9},
     {"_cpp11tesseract_print_params",                  (DL_FUNC) &_cpp11tesseract_print_params,                  1},
     {"_cpp11tesseract_tesseract_config",              (DL_FUNC) &_cpp11tesseract_tesseract_config,              0},
     {"_cpp11tesseract_tesseract_engine_internal",     (DL_FUNC) &_cpp11tesseract_tesseract_engine_internal,     5},

src/cpp11tesseract.cpp

@@ -7,11 +7,6 @@
 #define GenericVector std::vector
 #endif
 
-#include <poppler-document.h>
-#include <poppler-page.h>
-#include <poppler-image.h>
-#include <poppler-page-renderer.h>
-
 #include <memory>
 #include <list>
 #include <string>
@@ -297,58 +292,3 @@ data_frame ocr_data_internal(tesseract::TessBaseAPI *api, Pix *image) {
   if (!image) throw std::runtime_error("Failed to read image");
   return ocr_data_internal(api, image);
 }
-
-[[cpp11::register]] int n_pages(const std::string &file_path,
-                                const std::string &opw,
-                                const std::string &upw) {
-  auto doc = poppler::document::load_from_file(file_path, opw, upw);
-
-  if (!doc) {
-    throw std::runtime_error("PDF parsing failure.");
-  }
-
-  if (doc->is_locked()) {
-    throw std::runtime_error("PDF file is locked. Invalid password?");
-  }
-
-  return doc->pages();
-}
-
-[[cpp11::register]] list get_poppler_config() {
-  bool render_feature = poppler::page_renderer::can_render();
-  std::vector<std::string> formats = poppler::image::supported_image_formats();
-  writable::strings formats2(formats.size());
-  for (size_t i = 0; i < formats.size(); ++i) {
-    formats2[i] = formats[i];
-  }
-  return writable::list({
-    "render"_nm = render_feature,
-    "format"_nm = formats2
-  });
-}
-
-[[cpp11::register]] std::vector<std::string> poppler_convert(
-                      const std::string &file_path,
-                      const std::string &format, const std::vector<int> &pages,
-                      const std::vector<std::string> &names, const double &dpi,
-                      const std::string & opw, const std::string &upw,
-                      const bool &antialiasing, const bool &text_antialiasing) {
-  auto doc = poppler::document::load_from_file(file_path, opw, upw);
-  for(size_t i = 0; i < pages.size(); i++){
-    int pagenum = pages[i];
-    std::string filename = names[i];
-    std::unique_ptr<poppler::page> p(doc->create_page(pagenum - 1));
-    if(!p)
-      throw std::runtime_error("Invalid page.");
-    poppler::page_renderer pr;
-    pr.set_render_hint(poppler::page_renderer::antialiasing, antialiasing);
-    pr.set_render_hint(poppler::page_renderer::text_antialiasing,
-                       text_antialiasing);
-    poppler::image img = pr.render_page(p.get(), dpi, dpi);
-    if(!img.is_valid())
-      throw std::runtime_error("PDF rendering failure.");
-    if(!img.save(filename, format, dpi))
-      throw std::runtime_error("Failed to save file" + filename);
-  }
-  return names;
-}

tests/testthat/test-download.R

@@ -1,13 +1,13 @@
 test_that("tesseract_download works", {
+  skip_on_cran()
   expect_no_error(tesseract_download("ron", datapath = tempdir(), model = "fast"))
   expect_error(tesseract_download("rou", datapath = tempdir(), model = "fast"))
 })
 
 test_that("tesseract_contributed_download works", {
-
+  skip_on_cran()
   expect_warning(tesseract_contributed_download("grc_hist", datapath = tempdir(), model = "fast"))
 
-  # if OS is Linux
   if (identical(.Platform$OS.type, "unix") && !identical(Sys.info()[["sysname"]], "Darwin")) {
     expect_warning(warn_on_linux())
   }

tests/testthat/test-multipage.R

@@ -121,30 +121,18 @@ text <- input %>%
 cat(text)
 ```
 
-## Read from scanned PDF files and multipage TIFF files
+## Read from PDF files
 
-If your images are stored in PDF files they first need to be converted to a proper image format. Use a high DPI to keep quality of the image.
+If your images are stored in PDF files they first need to be converted to a proper image format. We can do this in R using the `pdf_convert` function from the pdftools package. Use a high DPI to keep quality of the image.
 
-```{r, eval=require(magick)}
+```{r, eval=require(cpp11poppler)}
+library(cpp11poppler)
 file <- system.file("examples", "ocrscan.pdf", package = "cpp11tesseract")
-text <- ocr(file)
+pngfile <- pdf_convert(file, dpi = 600)
+text <- ocr(pngfile)
 cat(text)
 ```
 
-The `ocr()` function can also extract text from multipage scanned PDF and TIFF files.
-The function will return a vector with the text found in each page. This is a good example
-that OCR is not perfect.
-
-```{r}
-file1 <- system.file("examples", "bondargentina.pdf", package = "cpp11tesseract")
-ocr1 <- ocr(file1, engine = eng)
-cat(ocr1)
-
-file2 <- system.file("examples", "bondargentina.tiff", package = "cpp11tesseract")
-ocr2 <- ocr(file2, engine = eng)
-cat(ocr2)
-```
-
 ## Tesseract Control Parameters
 
 Tesseract supports hundreds of "control parameters" which alter the OCR engine. Use `tesseract_params()` to list all parameters with their default value and a brief description. It also has a handy `filter` argument to quickly find parameters that match a particular string.